The frequency security of islanded microgrids is addressed through a distributionally robust scheduling method that optimizes heterogeneous frequency regulation resources, including generation sources, storage systems, and loads. A linear approximation model for frequency regulation power is developed by considering differences in response rates and dead zones. Frequency constraints under coordinated regulation are established and incorporated into the microgrid scheduling model. Wind power uncertainty is described using a moment-based ellipsoidal uncertainty set, and joint chance constraints are formulated to limit wind power output and frequency regulation reserves. Auxiliary variables are introduced to transform the joint chance constraints into individual chance constraints. Non-convex individual chance constraints are further converted by accounting for distribution unimodality. Bilinear terms are convexified using the McCormick envelope, reformulating the model into a mixed-integer second-order cone programming problem. Case studies demonstrate that the proposed method ensures the frequency of the islanded microgrid remains above 49.2 Hz following active power disturbances. The inclusion of frequency support from demand-side resources improves operational economics.